Experimental Selection of Machine learning Techniques and Image features to Detect “Cactus” Diseases
Hailay Beyene1, Narayan A.Joshi.2
1Hailay Beyene*. lecturer department of Computer Science Aksum University, Ethiopia
2Dr. Narayan A Joshi, Professor & Head, Dharmsinh Desai University, Nadiad.
Manuscript received on February 06, 2020. | Revised Manuscript received on February 10, 2020. | Manuscript published on February 30, 2020. | PP: 1438-1447 | Volume-9 Issue-3, February, 2020. | Retrieval Number: C5029029320/2020©BEIESP | DOI: 10.35940/ijeat.C5029.029320
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: Image is a very important data in machine learning. In order to select better features, feature extraction techniques and classifiers, intensive experiments are taken place using data. In this work, best feature, feature extraction technique and machine learning classifier are selected experimentally. Hence, bag of features were the best features experimentally out of color, texture and bag of features. Of color histogram, bag of features and GLCM (Gray-level co-occurrence matrix), bag of features extraction technique is found to be the best one experimentally. Of the machine learning classifiers shown in the scatter plot and confusion matrix, linear support vector machine is selected and the achieved accuracy is 97.2%.
Keywords: Cactus, bag of features, GLCM, Color histogram, Confusion matrix